A startup-focused adaptation of the Business Model Canvas. Trades partners, activities, and resources for Problem, Solution, Key Metrics, and Unfair Advantage.
Layout
Lean Canvas
9 blocks · 3-row grid
"Love the problem, not your solution."
What it is
A startup-stage canvas. Where the BMC asks "what does our business look like?", the Lean Canvas asks "what are we assuming, and what would change our minds?" Each block is a hypothesis you can test in a week. The discipline is to fill the riskiest assumptions first — usually Problem, Customer Segments, and Unique Value Proposition.
Origin
The Lean Canvas emerged from the lean-startup movement and was refined across multiple editions of community feedback. It explicitly trades four BMC blocks — Key Partners, Key Activities, Key Resources, Customer Relationships — for four startup-stage blocks: Problem, Solution, Key Metrics, and Unfair Advantage. The point is not academic completeness; it is rapid hypothesis testing.
When to reach for it
You are pre-product or pre-traction and need to focus the team on the right risks.
You are evaluating an idea quickly — fill out one canvas per candidate idea before picking which to pursue.
You are running a discovery sprint and want a shared artifact the whole team can challenge in 90 minutes.
The blocks
Worked example uses Airbnb in 2008 (then "AirBed & Breakfast").
What good looks like
List the top 1–3 problems your segment actually has. Bonus: list the existing alternatives they use today — that is the real competition, not the company on the next slide.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
Conference attendees in cities with sold-out hotels have no affordable last-minute lodging. Locals with a spare room have no easy way to monetise it. Existing alternatives — Craigslist sublets, Couchsurfing — feel unsafe and have no payment guarantee.
What good looks like
One sentence per problem. Resist the urge to describe features; describe the outcome the user gets.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
A simple website where locals list a spare air bed or couch for the duration of a specific event, and visitors book it up front with a credit card. Founders personally vet the first hosts.
What good looks like
One or two leading indicators that prove the model is working — activation rate, weekly retention, time-to-value. Vanity metrics (signups, traffic) don't count.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
Bookings per launched event, host-to-guest conversion (% of listed beds that got booked), and the week-2 repeat-host rate. Page views and signups were treated as inputs, not outcomes.
What good looks like
A single, clear, compelling message in plain English. The "high-concept pitch" (X for Y) goes underneath, as scaffolding, not as the headline.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
"Book a room with a local instead of a hotel" — for conferences where hotels are sold out and Craigslist feels sketchy.
What good looks like
Something competitors cannot easily copy or buy — proprietary data, an unfair distribution channel, a personal authority, a regulatory licence. If you can't name one yet, write "TBD" and treat that as a risk to retire.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
Founders embedded in the design and tech community (instant launch distribution at SXSW and the DNC), and a hand-curated trust layer where the team personally vetted the first hundred hosts.
What good looks like
Free + paid, inbound + outbound. The riskiest blank space is usually the channel that scales without a sales team.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
Event-specific landing pages tied to conferences (SXSW, the Democratic National Convention), founder cold-DMs to Craigslist hosts asking them to cross-list, and design-community blog coverage.
What good looks like
Be brutal about specificity. "Indie iOS developers shipping their first paid app" beats "developers." Sub-segment your early adopters explicitly.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
Two-sided. Supply: cash-strapped urbanites with a spare room during a big-city event. Demand: conference attendees priced out of hotels.
What good looks like
Just enough numbers to see whether the model breaks at small scale. Customer acquisition cost goes here, not in Revenue.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
Founder time, modest hosting, payment-processing fees, and famously the cereal-box hustle to cover rent — no paid marketing.
What good looks like
How much, how often, from whom. If the answer is "freemium," you also need a column for what triggers the upgrade.
Example — Airbnb in 2008 (then "AirBed & Breakfast")
A ~10% take rate on each booking, collected up front from the guest and paid out to the host after check-in.
How to use it
Start with Customer Segments and Problem in parallel — neither makes sense without the other.
Write the Unique Value Proposition next, then the Solution. The order matters: a solution without a UVP is technology in search of a problem.
Add Channels, Revenue Streams, and Cost Structure to check unit economics. Sketch the numbers, don't spreadsheet them.
Finish with Key Metrics and Unfair Advantage — what you'll measure to know it's working, and what stops a copycat from eating your lunch.
Common mistakes
Writing the Solution first, then back-solving Problem and Segments to match it.
Using vanity metrics — page views, signups — instead of the 1–2 numbers that prove the model.
Skipping Unfair Advantage. "We move faster" is not an advantage.
Spin up the canvas in one click. Copilot will score every cell against the same rubric this guide describes.
Keep reading